We had a GTK+ BOF today (2004-06-29) at GUADEC. I don't have complete list of people there. There were about 30-40 people total, with a good set of frequent contributors (Tor, Matthias, Soeren, Anders, Jonathan, Manish, and others that I'm forgetting) The first topic of discussion was GTK+-2.6; trying to finalize a rough feature set. Some of the possible features discussed were: Command line parser; Anders seems to have this pretty close to completion. The big remaining question is how we do the gnome_init() function of integrating command line arguments from various libraries into the parsing. The consensus we came to was perhaps a surprising one; libraries shouldn't have command line arguments, except perhaps one or two exceptions like --display. Regular expressions; Matthias's stuff based on PCRE still isn't quite ready to go in. It needs some actual use, and it also needs to be modified to integrate with the Glib UTF-8 functions in a more comprehensive way; it can't, for instance, handle case-insensitivity beyond ASCII. The current idea is to land it in libegg for now. Icon list. The main open question here is whether to use GtkTreeModel as the backend. Jonathan and Anders got together after the meeting to take a look at this and decided that it was not entirely trivial, but feasible. Anders took an action at the meeting to write up the API and status of EggIconList and send it to gtk-devel-list. Rotated text and labels should be simple if Owen finishes up a few more things for Pango-1.6. Matthias took an action to look again at his GtkAbout dialog work with the goals of a) keeping it very simple; if you want any more you can write it yourself b) forgetting about the single function call g_object_new() construction and just having some setters. Various picker buttons / entries were discussed: File/Directory picker. There was some considerably discussion of whether such a beast was an actually a good idea; it seems that at least *some* apps do genuinely need file and directory selection, and we're better off providing a widget to do it well than have everybody do it themselves badly. James Cape was nominated in absentia to write up his work as a proposal and send it to gtk-devel-list Date/Time. Seems useful, not clear if we have a close enough prototype to get something in for 2.6. During this discussion the ancient "Time API to go with date API" bug (http://bugzilla.gnome.org/show_bug.cgi?id=50076) came up. Apparently JP Rosevear has strong feeling about some capabilities in this area that should be in GLib, so he was encouraged in absentia to write this up for gtk-devel-list. The next two topics were GTK+-2.8 plans. Didn't really have any time to discuss this. See http://people.redhat.com/otaylor/guadec5/ for discussion of Cairo plans. How to deal with the big accumulating pile of bugs. The idea of releases targeted at particular areas (resolve/wontfix *all* GtkNotebook related bugs) came up again. Should this targeting be for a stable release or an unstable release? There's a good argument for doing it for unstable releases ... some of the bugs are going to be API bugs. But you'd want to apply relevant patches to the stable branch at the same time. And as a corollary to the last, the question of when the first unstable release of GTK+-2.5 should be made came up. "Almost immediately" was the thought. Regards, Owen
Attachment:
signature.asc
Description: This is a digitally signed message part